Different Speeds, Different Strategies: Mapping AI Adoption Across Career Areas to Prioritize Workforce Preparation

[00:00:00] What if I told you that the decline in jobs that are supposedly being wiped out by AI actually started months before ChatGPT even existed? I mean, it completely shatters the narrative we hear every single day. The data actually shows that the drop in demand for these highly vulnerable roles was, well, it was well underway before generative AI even became a household name. Welcome to this deep dive. If you've been, you know, anywhere near the internet or television lately, you know the narrative.

[00:00:29] It's this terrifying metaphor of a massive unavoidable tsunami, like a towering wall of artificial intelligence coming for every single job and every single industry all at once. Yeah, it's everywhere. Right. And you sitting there listening are probably wondering how to brace for impact without just completely drowning in information overload.

[00:00:48] So today, our mission is to look at the hard actual job market data from the first half of 2026 to see if that tsunami is, well, if it's actually real. Mm-hmm. We're pulling from a brand new academic article by Jonathan H. Westover. It's published in the Human Capital Leadership Review and it's titled Different Speeds, Different Strategies. And Westover's research is honestly, it's a breath of fresh air. It forces us to abandon all the hype and look really closely at reality.

[00:01:17] He uses labor market data from Lightcast, which is this analytics firm that tracks millions of actual U.S. job postings, to figure out what employers are genuinely paying for today. Right. And spoiler alert, the uniform tsunami, it just doesn't exist. Okay, let's unpack this. We really have to figure out why everyone is so panicked in the first place because it comes down to a fundamental misunderstanding of what job data is actually telling us, right?

[00:01:42] The headlines constantly swap two crucial concepts, exposure and adoption. Yeah, confusing those two words is the entire root of the panic. Let's look at exposure first. Exposure is strictly the hypothetical potential for an AI tool to perform or assist with a specific task in a given job. Okay.

[00:02:03] So when you see a terrifying headline screaming that like 80% of the U.S. workforce could have at least 10% of their tasks affected by AI, you are looking at an exposure metric. Just potential. Exactly. It's a measure of what the technology is theoretically capable of doing in a complete vacuum. So it's like a weather forecast warning that your town has the geographical potential to flood. That's exposure. We are all out here panicking about a hypothetical forecast. That's a perfect way to put it.

[00:02:32] Adoption, however, is what bosses are actually doing. It is realized organizational demand. Okay. What does that mean exactly? It's the hard evidence of employers actively requiring AI skills in their current job postings and, you know, paying salaries for those skills. The article points to a study by economists at the Federal Reserve Bank of New York to highlight how dangerous it is to conflate the two. Right. They examined the career paths with the absolute highest AI exposure.

[00:02:59] And they did find a decline in job postings for those highly exposed roles. But wait, that's the stat you mentioned at the very beginning, right? That relative decline started before the AI boom even kicked off in late 2022. Precisely. The New York Fed found that the dip was driven by broader economic restructuring, you know, shifts in remote work, interest rate changes, not an AI takeover. Oh, wow. Yeah. In fact, they found very little evidence of any distinct AI driven decline in labor demand so far.

[00:03:29] And even more telling when they surveyed the firms directly, employers overwhelmingly stated they were focused on retraining their current workers to use these new tools rather than, you know, firing them and replacing them with algorithms. So the weather forecast says massive flood. But when we actually look at the water levels in our neighborhoods, it's just a puddle. Pretty much. But if the tsunami forecast is wrong, what does the actual map look like? How is AI really moving through the economy right now?

[00:03:57] Well, this is where the light cast job posting data becomes incredibly useful. Westover plots every major career area on a grid based on two specific metrics. The horizontal axis measures current AI adoption. So what percentage of job postings in a specific field require AI skills today? Okay. Got it. And the vertical axis measures the growth rate, basically how much that specific demand has spiked compared to the previous year, 2025. And when you map it out that way, the massive uniform wave disappears entirely.

[00:04:28] Instead, careers clump into four very distinct climates. Finding out which climate your career sits in is, well, it's the key to figuring out your personal strategy. Absolutely. So the first climate is the AI hotspot. High current adoption combined with blistering fast growth. We are talking about human resources, design, media and writing. The storm is truly here for these professionals, isn't it? It is.

[00:04:54] And the financial incentives driving these hotspots are just staggering. According to the data, job postings that require AI skills in these areas carry an average salary premium of roughly $18,000 right now. Wait, really? Yes. 18,000? Yes. That translates to an almost 28% bump in pay just for having those specific skills on your resume. It reflects a desperate scarcity, really.

[00:05:20] Companies know they need AI proficiency in design and media, but they just can't find enough people who actually know how to apply it practically. I mean, a 28% bump is life-changing money. That tells you everything you need to know about where the immediate demand is. Definitely. But let's look at the second climate, the emerging frontiers. This one is fascinating because the current adoption is actually very low, but the growth rate is astronomical. Right. This includes customer support, finance and clerical work.

[00:05:50] So, AI isn't a requirement in every job posting yet, but it is sneaking up incredibly fast. It is. And the way the light cast chart visualizes this is vital to understand. So, the horizontal axis, the current adoption rate, it's plotted on a logarithmic scale. Oh, okay. You have to translate that for us. What does a logarithmic scale actually mean for someone sitting in an emerging frontier like finance? Oh, I love this part.

[00:06:17] Think of it like the Richter scale for earthquakes or compounding interest in a bank account. Okay. On a standard chart, the distance between one and two is the exact same as the distance between 10 and 11. But on a logarithmic scale, the values multiply as you move to the right. Right, right. So, small visual distances on the left side of the chart, where the emerging frontiers are, represent much smaller absolute differences than the exact same visual jump on the right side.

[00:06:46] Meaning, a tiny little shift to the right on this chart represents an absolute explosion in real world job demand. Exactly. A single millimeter of movement on the left side of that chart means thousands of new job postings suddenly requiring AI skills. It really illustrates how an emerging frontier can look quiet and unassuming one day and then just exponentially explode the next. Like a snowball rolling down a mountain. Yeah, it looks totally harmless at the top, but by the time it reaches the bottom, it's a full-blown avalanche.

[00:07:15] That is a terrifyingly fast transition. But what about the people who actually build the technology? Like software engineers, data scientists, IT professionals. You think they would be in the fastest growing hotspots, but they sit in the third climate, the established AI hubs. Which is high adoption, but surprisingly slower growth. Right. Why is that? Well, the slower growth isn't a sign of stagnation. It's actually a sign of maturity. I mean, machine learning, natural language processing, data science.

[00:07:45] These were foundational to IT and engineering long before generative AI became a mainstream buzzword. Oh, of course. Yeah. They've been using this stuff for years. Exactly. For software developers, AI is an evolution of their existing toolkit, not some sudden alien revolution. They already have a really high baseline of adoption, so they aren't experiencing that massive overnight spike in growth. Makes sense. And that leaves us with the fourth climate, the AI cold zones.

[00:08:13] Low current adoption and slower growth. This is healthcare, agriculture, construction, community services. Yeah. In these fields, physical presence, human empathy, and highly unpredictable physical environments are just the core of the work. You can't, you know, automate the physical act of framing a house or holding a patient's hand. But the danger here, and really the whole point of looking at this map, is the friction of misjudging your climate.

[00:08:39] What actually happens if a company, even just a worker, ignores this map and tries to do AI everywhere all at once or completely ignores it? It's risky. Let's examine the penalty for moving too slowly first. If you are operating in a hotspot or an emerging frontier and you adopt a wait-and-see approach, the data suggests you will be outpaced by competitors almost immediately. How fast are we talking? Fast. The text cites a pre-registered experiment involving professional writers.

[00:09:07] When they incorporated generative AI into their workflows, the time it took to complete tasks dropped by roughly 40%. 40%? That's huge. And the output quality, which was evaluated by independent judges, actually rose by 18%. Okay. I have to pause and push back on this a bit because this is where the anxiety lives for so many people, right?

[00:09:31] If I'm a writer or a marketer and this tool makes me 40% faster, the simple math says my boss only needs to employ roughly half as many writers to get the exact same output. Right. So, displacement just feels like the only logical outcome of that kind of efficiency. It seems like simple math. I totally get that. But labor economics rarely follows simple linear subtraction. The text addresses this tension using the historical arrival of the automated teller machine. The ATM. Yeah, the ATM.

[00:10:01] When ATMs were rolled out, the widespread assumption was that bank tellers would be entirely eradicated. I mean, a machine was doing their core task, handing out cash. But teller jobs didn't collapse. Really? No. ATMs dramatically lowered the operational cost of running a bank branch. And because it was so much cheaper, banks expanded aggressively, opening thousands of new branches. I mean, I get the ATM analogy, but I'm still hung up on the difference in the type of labor.

[00:10:28] An ATM is a mechanical box doing a mechanical task, just spitting out $20 bills. Sure. Writing, designing, strategizing. That is cognitive labor. It's thinking. If a machine can do the thinking, I don't just shift to a different task like a bank teller selling a mortgage. I feel like I become obsolete. The cognitive nature of the labor does change the dynamic. That's fair. But the economic principle, what economists call demand elasticity, remains very similar.

[00:10:57] When the cost of producing written content drops by 40 percent, a company doesn't usually say, well, let's produce the exact same amount of content and fire half the staff. They don't. Usually not. Instead, they realize content is now cheap. So they decide to produce 10 times as much content. They hyper personalize their marketing. They launch entirely new campaigns they previously couldn't afford. So they expand the pie. Exactly. The demand for the output explodes.

[00:11:22] And suddenly you need humans not just to generate the raw text, but to edit, strategize, ensure brand voice and manage this massive new volume of content. The nature of the writer's job shifts from staring at a blank page to curating and refining high level strategy. That makes a lot of sense. And in customer support, there's another study of over 5,000 agents that showed a 14 percent boost in issue resolution using AI and a staggering 34 percent boost for novice workers. Wow.

[00:11:49] If you move too slow, your competitor adopts this, scales their operations massively and crushes you on volume and response time. Okay. So the risk of moving too slow is very clear, but you mentioned a penalty for moving too fast as well. If AI is this powerful, why wouldn't the CEO just mandate it across every single department tomorrow? Because treating a cold zone like a hotspot triggers what economists call the productivity J curve. The J curve. Let's break down the mechanics of that. So imagine the letter J. It dips down before it goes up.

[00:12:19] When an organization adopts a massive general purpose technology like AI, overall performance almost always drops initially. Why does it drop? Because you can't just drop an AI chat bot onto an employee's desk and expect instant results. You have to completely restructure how the work flows. You have to rewrite standard operating procedures, audit your data security, and spend hundreds of hours training employees who frankly might be pretty resistant to the change.

[00:12:47] You are breaking their old habits, which takes time and creates massive friction. Exactly. It creates chaos. If a company forces that disruptive restructuring onto a department in a cold zone, like a department that doesn't actually have a high immediate demand for AI skills, they just plunge their workers into the bottom of that J curve dip for no immediate gain. Ouch. Yeah. They destroy productivity and create deep disillusionment.

[00:13:12] So a one size fits all approach is just a guaranteed path to either getting left behind by competitors or completely drowning your team in the J curve. We need tailored strategies for the real world. We really do. Let's translate this map into actionable steps for the listener, depending on their specific career climate. Let's start with the hot spots. The strategic posture here is act now, but it's not about generic company wide AI seminars. It requires highly specific role based sprints.

[00:13:42] Like what? You need to map out your daily workflow, identify the routine time consuming tasks and figure out how to offload them to AI. The source gives a really fascinating real world example of this from IBM actually. In 2025, their CEO publicly stated they had replaced hundreds of routine HR jobs with AI. Yes. And that is the scary headline everyone read. But the crucial context that was ignored is that overall employment at IBM actually increased.

[00:14:09] They took the massive savings from automating those routine administrative processes and reinvested it directly into hiring for roles that heavily required complex human judgment, client interaction and strategic thinking. That is the perfect blueprint for a hotspot. Automate the routine to fund the uniquely human. And what's fascinating here is if you find yourself in an emerging frontier like finance or customer support, the posture is invest early. Invest early. Yes.

[00:14:37] This is arguably the most lucrative quadrant on the map. Adoption is currently low, but the growth is exponential. If you learn these tools now, you get to shape how they actually enter your industry. And the case study for the frontier is Morgan Stanley. They didn't view AI as a way to replace their financial advisors. They built an internal AI assistant designed to instantly parse through thousands of pages of complex internal economic research and draft meeting notes.

[00:15:06] It was a tool built specifically to eat up the tedious prep work. Right. And by eating up the prep work, those human advisors suddenly had significantly more hours in the day to actually sit down, look their clients in the eye and build trust. They use the AI as an amplifier of human skill, not a substitute for it. I love that. Moving to the third climate, the established AI hubs like software engineering. The posture is manage evolution. Because they don't need a revolution. Exactly.

[00:15:35] They just need a capability refresh cycle. Exactly. Think about what a software engineer actually does. They solve complex logic puzzles. When GitHub Copilot was introduced, data showed it made developers 56% faster at writing code. Which is massive. It is. But it didn't fundamentally change the core nature of their job. They are still solving logic puzzles. They just have a much more sophisticated tool to translate that logic into code.

[00:16:01] It's about updating your proficiency with the new tools, not reinventing your entire profession. Here's where it gets really interesting. Which brings us to the cold zones. Healthcare, agriculture, construction. The posture here is build readiness. And this is where people get confused, I think. They assume a lack of AI job postings means AI doesn't exist in these fields at all. Right. But the innovation is happening. Right. It's just focused on entirely different pinpoints.

[00:16:30] The example in the text from the Permanente Medical Group is brilliant. They rolled out ambient AI scribes to 10,000 physicians. Ambient scribes. Yeah. It's essentially a secure microphone in the exam room. It listens to the natural conversation between the doctor and the patient. And it automatically drafts the clinical notes and paperwork. That's incredible. The doctor still reviews and signs off on every single word. So the human medical judgment is entirely preserved.

[00:16:58] But it removes hours of soul-crushing administrative burnout at the end of the day. Exactly. In a cold zone like healthcare, the AI isn't taking over the core job. It's actually being used to protect the most inherently human part of the job, the doctor-patient connection, by completely absorbing the paperwork. You are building readiness by getting the workforce comfortable with the technology in a low-risk, highly helpful way.

[00:17:24] And that deliberate readiness ensures that when more complex diagnostic AI tools eventually do reach the healthcare sector, the workforce is already culturally and technically prepared to integrate them. So we've covered every climate on the map, but the article introduces a wild card. There is one specific sector currently sitting down in the cold zone that actually controls the fate of every other career we just talked about. Ah, yes. And that is education. Education is entirely unique.

[00:17:51] On the Lightcast map, its own internal AI adoption is low and the growth is relatively slow. But Westover labels education as a downstream multiplier. A downstream multiplier. Let's break that down. Why does education carry so much weight? Because the education sector is responsible for training the human beings who will eventually walk into the hotspots and the frontiers. If high schools and universities fail to figure out how to integrate AI effectively, the entire workforce pipeline basically collapses.

[00:18:21] Oh, wow. Yeah. The downstream companies won't have anyone capable of doing those high-level human judgment tasks. And according to a study highlighted in the text, if we implement AI poorly in education, the results are actually destructive. They looked at high school math students in Turkey. Right. The researchers split the students into two groups. The first group was given an unrestricted general purpose AI to help them practice math problems. They could ask it whatever they wanted.

[00:18:47] The second group was given a guard-railed AI, essentially an AI program to act like a coach. It would only offer hints, never the final answer. The contrast in the outcomes is really profound here. During the practice sessions, the students with the unrestricted AI seemed to be doing great. They were getting the right answers quickly. Of course.

[00:19:06] But when both groups were later given a traditional exam without any AI assistance, the students who used the unrestricted AI actually performed significantly worse than if they had just studied normally with no technology at all. Because they never actually engaged in the productive struggle of learning. They bypassed the thinking entirely. Meanwhile, the students with the guard-railed AI, the coach instead of the answer key, were forced to use their own reasoning. They received just enough help to avoid getting completely stuck.

[00:19:35] And they completely avoided the cognitive harm seen in the other group. I mean, this raises a massive red flag. If we just hand learners these incredibly powerful tools without any kind of structural guardrails, we aren't creating a more capable, tech-savvy workforce. We are creating a workforce that is entirely dependent on a crutch. Exactly. If education is the downstream multiplier and we multiply a deficit in critical thinking, that deficit bleeds into every single company in every single industry.

[00:20:03] It is arguably the most critical warning in the entire paper. We have to teach humans how to think with the tool, not how to let the tool think for them. So, what does this all mean for you? The overarching takeaway from Westover's research is that you can officially stop bracing for a massive, uniform AI tsunami. It just isn't real. No, it's not. Instead, look at the flood map. Identify exactly which climate your specific career is in today.

[00:20:33] Are you in a hot spot where you need to act immediately? A frontier where you should invest early? A hub managing evolution or a cold zone building readiness? Stop confusing theoretical exposure with actual job displacement. Tailor your strategy to your specific climate and focus relentlessly on building a foundation that combines technical proficiency with the deep human judgment that machines still cannot replicate. Which leaves us with one final provocative thought to chew on before we go.

[00:21:04] Earlier, we discussed how companies in those emerging frontiers are using AI to capture the tacit knowledge of their absolute best, top performing human experts. Right. They monitor the best customer service agents, train the AI on their specific behaviors, and use that model to bring the novices up to speed. It's incredibly smart for optimizing a business today. But think about the long term implication of that. That's a big question.

[00:21:27] What happens in a decade or two when those original human experts, the ones who learned the hard way through productive struggle before AI even existed, eventually retire? If the entire next generation of workers is trained exclusively using AI guardrails and hints based on past data, where will the new human ingenuity actually come from? Yeah. Who will generate the breakthrough ideas required to train the AI of tomorrow? Will we eventually hit a ceiling where our systems just recycle the expertise of the past?

[00:21:55] Or will human creativity, as it always does, somehow find an entirely new frontier to master? That is the defining challenge for the next generation of workers. Something to think about the next time someone tries to sell you on the tsunami. Thanks for joining us on this deep dive. We'll catch you next time.